hi @KOLANICH

Your two issues are unexpected. Can you provide one tiny program example for each of them, so I can check ?

1

import scipy

2 try to launch jupyter notebook

hum,

It works in "full" winpython:

Python 3.6.1 (v3.6.1:69c0db5, Mar 21 2017, 18:41:36) [MSC v.1900 64 bit (AMD64)] on win32
Type "copyright", "credits" or "license()" for more information.
>>> import scipy
>>> 

Would it be possible you are using WinPython-32bit-3.6.1.0Zero ? it does not contain scipy, nor jupyter.
it contains only https://github.com/winpython/winpython/blob/master/changelogs/WinPythonZero-3.6.1.0.md

Concerning Windows 32bit, support from new "numeric" packages seems to go down since end of 2016. It may die by 2020 (a much slower death than expected, as first amd64 was in 2003, 14 years ago)

suggestion, try by order of workarounds or preferences :

• in "winpython-32bit-3.6.1.1\python-3.6.1.amd64", rename "python_onHold._pth" as "python._pth"
• try with a winpython 3.5,
• try with a winpython 3.4 version, (to try first, if you are not on windows 10),
• try with anaconda or enthought free distribution

try with anaconda

In fact it is the quickest and the most effortless solution. I have done exactly this. Spent an hour installing that, but that worth: everything worked out of the box. Unfortunately on another PC (the one with XP, python 3.4, no Visual Studio) I cannot use Anaconda, the troubles are the same. IMHO there is something wrong with the releases.

On windows XP, try the last WinPython that was release just before XP support ended from Microsoft. After that moment, a lot of binary packages started to stop working

i think i could have face the same problem too after i installed/update tensorflow which update the numpy due to requirement. if i remember correctly, it was fine before tensorflow installation. still trying to figure it out how to support both for my learning.

Traceback (most recent call last):
File "Part2.py", line 21, in
from sklearn.datasets import load_iris
File "E:\WinPython\WinPython-64bit-3.5.3.1Qt5\python-3.5.3.amd64\lib\site-packages\sklearn_init_.py", line 57, in
from .base import clone
File "E:\WinPython\WinPython-64bit-3.5.3.1Qt5\python-3.5.3.amd64\lib\site-packages\sklearn\base.py", line 10, in
from scipy import sparse
File "E:\WinPython\WinPython-64bit-3.5.3.1Qt5\python-3.5.3.amd64\lib\site-packages\scipy_init_.py", line 61, in
from numpy._distributor_init import NUMPY_MKL # requires numpy+mkl
ImportError: cannot import name 'NUMPY_MKL'

E:\WinPython\WinPython-64bit-3.5.3.1Qt5\scripts>python
Python 3.5.3 (v3.5.3:1880cb95a742, Jan 16 2017, 16:02:32) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.

import numpy
numpy.version.version
'1.12.1'
exit()

E:\WinPython\WinPython-64bit-3.5.3.1Qt5\scripts>

import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
hello = tf.constant('Hello, TensorFlow!')
sess = tf.Session()
print(sess.run(hello))
b'Hello, TensorFlow!'
print (tf.version)
1.1.0
import scipy
Traceback (most recent call last):
File "", line 1, in
File "E:\WinPython\WinPython-64bit-3.5.3.1Qt5\python-3.5.3.amd64\lib\site-packages\scipy_init_.py", line 61, in
from numpy._distributor_init import NUMPY_MKL # requires numpy+mkl
ImportError: cannot import name 'NUMPY_MKL'

from WinPython Control Panel.exe scipy is 0.19.0

another version WITHOUT tensorflow installation/upgrade is ok.
E:\WinPython\WinPython-64bit-3.6.1.0Qt5\scripts>python
Python 3.6.1 (v3.6.1:69c0db5, Mar 21 2017, 18:41:36) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.

import scipy
scipy.version
'0.19.0'
import numpy
numpy.version.version
'1.11.3'
import os
os.environ['TF_CPP_MIN_LOG_LEVEL']='2'
import tensorflow as tf
hello = tf.constant('Hello, TensorFlow!')
sess = tf.Session()
print(sess.run(hello))
b'Hello, TensorFlow!'
print (tf.version)
1.0.1

after some code rearrangement, we can see the config.py different (attached).

hopefully the next release will solve this issue.